Key takeaways
- Attribution assigns credit for a conversion across the touchpoints that influenced it.
- Every model is a simplification; none is objectively correct.
- Last-click over-credits the final touch; first-click over-credits discovery.
- Platform-reported conversions overlap and double-count across channels.
- Pair a directional attribution model with blended metrics (MER) for budget decisions.
What is marketing attribution?
Attribution is the practice of deciding which marketing touchpoints get credit for a sale. A single purchase might involve a Meta ad, a later Google search, an email, and a direct visit. Attribution models distribute the credit among them — and that distribution drives how you judge each channel and where you put budget.
There is no perfectly true attribution
Customer journeys are messy and partly invisible. Attribution is a useful approximation, not ground truth — treat every model's output as directional.
What are the main attribution models?
| Model | Who gets credit | Bias |
|---|---|---|
| Last-click | The final touch before purchase | Over-credits bottom-funnel and branded search |
| First-click | The first touch | Over-credits discovery, ignores closers |
| Linear | All touches equally | Treats a minor touch like a decisive one |
| Time-decay | Recent touches more | Still discounts important early discovery |
| Data-driven | Algorithmic, by contribution | Opaque; varies by platform and data volume |
Last-click remains the most common default because it's simple, but it systematically flatters channels that appear at the end of the journey — branded search and retargeting — while undervaluing the prospecting that created demand in the first place.
Where does attribution mislead you?
- Double-counting: each platform claims the same conversion under its own model.
- View-through inflation: counting sales after an ad was merely seen, not clicked.
- Walled gardens: Meta and Google can't see each other's touches, so both over-credit themselves.
- Privacy and tracking loss: blocked cookies and limited tracking leave gaps that models fill with assumptions.
Don't reallocate budget on one model alone
Cutting a prospecting channel because last-click shows low ROAS can collapse the demand feeding your high-ROAS branded search. Cross-check with blended performance.
How does GA4 fit in?
GA4 is a web-analytics tool: it models sessions, channels, and conversions on your site, with its own attribution settings. It's valuable for understanding on-site behavior and channel mix, but it is not the same as your store's revenue ledger, and it won't perfectly match what Shopify or the ad platforms report.
Use GA4 to understand how channels contribute to traffic and engagement, and use your commerce platform as the source of truth for revenue. Expecting them to reconcile to the dollar leads to endless, fruitless debugging.
Why the numbers never perfectly matchRead: GA4 vs ad-platform revenueHow should you actually use attribution?
- 1
Pick one directional model
Choose a model and stick with it for consistency, knowing its bias.
- 2
Anchor on blended metrics
Use blended ROAS / MER (total revenue ÷ total spend) for budget-level decisions that can't be double-counted.
- 3
Watch incrementality signals
When you change spend on a channel, watch what happens to total revenue, not just that channel's attributed revenue.
- 4
Use GA4 for behavior, store for revenue
Let each tool do its job instead of forcing them to agree.
Frequently asked questions
Which attribution model is best for ecommerce?
There's no single best model — each has a built-in bias. Many teams use last-click for consistency while anchoring budget decisions on blended metrics like MER. The model matters less than understanding its bias and not over-trusting it.
Why do Meta, Google, and GA4 all report different conversions?
Each uses its own attribution windows, models, and tracking, and the platforms can't see each other's touchpoints. They will always disagree. Use them directionally and rely on your commerce platform for true revenue.
Is blended ROAS a form of attribution?
It's the opposite — blended ROAS deliberately ignores attribution by dividing total revenue by total spend. That's its strength for budget decisions: no touchpoint can be double-counted, so it can't be gamed by attribution choices.
Keep reading